TREC 2025 Proceedings

4method_merge

Submission Details

Organization
UTokyo
Track
Retrieval-Augmented Generation
Task
Retrieval Only Task
Date
2025-08-16

Run Description

Is this a manual (human intervention) or automatic run?
automatic
Does this run leverage neural networks?
yes
Does this run leverage proprietary models in any step of the retrieval pipeline?
yes
Does this run leverage open-weight LLMs (> 5B parameters) in any step of the retrieval pipeline?
no
Does this run leverage smaller open-weight language models in any step of the retrieval pipeline?
yes
Was this run padded with results from a baseline run?
no
What would you categorize this run as?
Generation-in-the-loop Pipeline
Please provide a short description of this run
Comprehensive 4-method hybrid retrieval system combining two dense retrievers (Qwen3-0.6B and BGE-small-en-v1.5, both HyDE-enhanced with query:HyDE 0.3:0.7 weighting) and two sparse methods (SPLADE learned sparse representations and BM25 with GPT-4.1-generated keyword expansion). All four retrieval streams produce top-1000 results that are fused using Reciprocal Rank Fusion (RRF, k=60) to leverage diverse relevance signals. Final ranking performed by GPT-4.1-mini using sliding window reranking (window=10, stride=5, 3 passes) with enriched document context including title, URL, and segment content for improved relevance assessment.
Please give this run a priority for inclusion in manual assessments.
2

Evaluation Files

Paper